BRNet: Ballistocardiogram Artifact Removal in EEG Acquired Inside MRI via Deep Learning

Motivation: EEG recordings inside MRI scanners are corrupted by BCG artifacts, which obscure genuine neural activity. Goal(s): Develop BRNet, a deep learning method to effectively reduce BCG artifacts while preserving essential neural signals. Approach: BRNet uses ECG input to estimate and remove BCG artifacts from EEG data using a 1D U-Net architecture, compared to the OBS method. Results: BRNet significantly reduces BCG artifacts and preserves alpha band differences between eyes-open and eyes-closed states, outperforming OBS. Impact: Enhances EEG data quality in MRI environments, improving reliability in neuroscience research and clinical diagnostics.